The closest I'm aware of is this: http://papers.nips.cc/paper/6650-toward-multimodal-image-to-image-translation http://papers.nips.cc/paper/6650-toward-multimod

The closest I'm aware of is this: http://papers.nips.cc/paper/6650-toward-multimodal-image-to-image-translation http://papers.nips.cc/paper/6650-toward-multimodal-image-to-... However, it requires rectified, segmented training data, unlike the above.